Securing the Future Regional and National Programs to Support the Semiconductor Industry (2003) / Chapter Skim
Currently Skimming:
SEMATECH Revisited: Assessing Consortium Impacts on Semiconductor Industry R&D
SEMATECH Revisited: Assessing Consortium Impacts on Semiconductor Industry R&D
Pages 254-282
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 254... ...
SEMATECH's cooperative structure to some extent was stimulated by Japan's VLSI projects of the late 1970s, which were perceived in the United States (and in Japan, for that matter) as having greatly advanced the technological competence of Japanese semiconductor producers.2 A 1987 Defense Science Board report pointing to deterioration in the relative position of U.S.
|
From page 255... ...
semiconductor materials and equipment producers formed a complementary organization, SEMI/SEMATECH, in 1987 specifically for the purpose of cooperating with SEMATECH. SEMI/SEMATECH was granted SEMATECH membership and a seat on the SEMATECH board of directors, and became the official vehicle for the organization of SEMATECH development projects and the teaming of materials and equipment producers.
|
From page 256... ...
7The cost structure of SEMATECH membership does not appear to be publicly reported. It is known that the originally approved cost structure consisted of 1 percent of a company's semiconductor revenues, with a minimum of $1 million, up to a maximum of 15 percent of the privately contributed portion of the SEMATECH budget ($100 million through the mid-199Os)
|
From page 257... ...
It was also certainly perceived as a major force in Japan, where the SEMATECH model greatly influenced the formation of a new generation of semiconductor industry R&D consortia in the mid-1990s.~2 Despite these major changes there has been no new research on the impact of SEMATECH since these two early studies and little discussion of the limitations in the Irwin and Klenow study, the only one to date to seek to measure the impact of SEMATECH on semiconductor industry R&D. This paper attempts to update this early analysis and remedy some of its problems.
|
From page 258... ...
If we call parameter ~ a "spillover coefficient," and assume R&D efforts among different firms are perfect substitutes, then Figure 1 shows an R&D cartel as the institutional realization of pure coordination. Without coordination firms 1 · Picture of an R&D cartel Competition: H1, H2 chosen to maximize individual Cartel: H1, H2 chosen to maximize joint L1 H1 Eff.
|
From page 259... ...
In this latter case the story includes both coordination and information sharing, and the spillover coefficient ~ for another firm's R&D investment is always 1, since it is performed by the joint venture and given to both firms. If the joint venture is somehow less efficient than individual-firm R&D efforts, then ~ can be less than 1 and the possible values for the spillover coefficient are not materially different than in the case of the pure R&D cartel.
|
From page 260... ...
yields the most "effective R&D."~3 In short, if we argue that SEMATECH is like an R&D joint venture in which firms sit down together and jointly figure out which R&D investments are worth making in order to maximize the entire industry's profitability, then agree to fund those investments, then SEMATECH must increase the level of "effective R&D" being performed in the industry. Conversely, if SEMATECH were like an R&D joint venture in which members voluntarily decide how much to contribute, if they wish, with the ability to "free ride" on the contributions of others a real option, then SEMATECH would decrease the overall level of "effective R&D" performed in the industry.
|
From page 261... ...
Even more tenuous is the Irwin and Klenow claim that a government subsidy is clearly justified in order to stimulate industry commitment to R&D if externalities are present and completely unjustified if information sharing is the exclusive motivation for cooperation.~5 Kenneth Arrow in his seminal 1962 contribution i4"Recall the predicted impact of SEMATECH ranged from -~* of sales under the sharing hypothesis to +1 percent of sales under the commitment hypothesis.
|
From page 262... ...
If this is not the case if they are imperfect substitutes or even complements the logic falls apart. For example, Irwin and Klenow argue that if eliminating duplication in R&D information sharing is the sole motivation for cooperation, then a dollar of SEMATECH R&D simply replaces inframarginal investments in R&D within all member companies.
|
From page 263... ...
Finally, Irwin and Klenow argue that if pure information sharing is what SEMATECH is about and SEMATECH R&Dis a perfect substitute for internal R&D, the net impact of a SEMATECH effort of size S would be to cut every i8See Irwin and Klenow, op.
|
From page 264... ...
"Effective" R&D summed up over the entire industry could actually rise as new entrants added their efforts to the unchanged totals for incumbent firms. Our conclusion, as before, is that sweeping claims that estimated coefficients draw sharp lines in the sand between competing hypotheses about SEMATECH and the nature of its R&D projects are ill-founded.
|
From page 265... ...
Heuristically we can further argue that if we allow spillovers between firms from internal R&D or consortium R&D to be a complement to rather than a substitute for internal efforts, the expected decline in firm R&D to sales ratios associated with pure information sharing might actually be reversed and a net positive effect on R&D to sales ratios observed. A positive impact of SEMATECH would show that the "coordination" effects more than counterbalanced an expected negative "sharing" effect.
|
From page 266... ...
, another 27 were more narrowly focused IC manufacturers, 10 firms designed but did not manufacture ICs, l l firms manufactured simple discrete semiconductor devices, and the remaining 12 firms were not directly involved in semiconductor production. We have also extended the original sample of firms used by Irwin and Klenow from 1993 through 1998 and have added to the sample additional firms classified by Compustat as semiconductor producers, increasing sample size significantly.
|
From page 267... ...
but only six years in which any of the SEMATECH membership variables are nonzero. By increasing our sample length to 1998, we almost double the time dimension of the SEMATECH membership variables and, we hope, also reduce bias issues coming from small sample size.
|
From page 268... ...
The differenced model with the lagged it&D-to-sales ratio instrumented produced even larger point and longrun multiplier estimates, though standard errors were even larger. Note that all SEMATECH members belong to this group of "broad" IC producers.
|
From page 269... ...
Table 4 also shows that using current nominal assets (FW weights) as weights in lieu of real assets (asset weights deflated by an investment price index for the semiconductor industry, the original Irwin and Klenow procedure)
|
From page 270... ...
We note that we rejected the inclusion of discrete device producers and materials and equipment producers when appropriate Wald tests were run. New Data One strategy for improving these estimates is to extend and refine the original Irwin and Klenow sample.
|
From page 272... ...
semiconductor materials and equipment suppliers by adding selected firms from SIC code 3559 (a grab-bag category for semiconductor equipment suppliers) to the semiconductor equipment and materials suppliers included in the original Irwin and Klenow sample, and then coding a dummy variable to reflect participation in the SEMI/SEMATECH supplier consortium for a company in a given year.
|
From page 273... ...
273 ·_4 an an by o o so To Cal an ED o Cq an .= an VO s~ ,~o.
|
From page 274... ...
Estimated standard errors continued to be large, and parameter estimates were not particularly stable when different variants of the same basic estimators were run over different subsets of companies and periods. Our conclusion is that the basic model specification used here is not capturing important aspects of the data well.
|
From page 275... ...
However, we have taken pains to point out this cannot be interpreted as showing anything about the nature of the effort going on within SEMATECH, or as proving whether the effort justified a government subsidy. Finally, the underlying models of R&D cooperation that ultimately must be the basis of a scientific effort to untangle the chains of causality are simply too simplified at this point to capture the complexity of the real world of SEMATECH: a real world in which companies committed to R&D carried out within a joint venture while at the same time competing through internal R&D efforts that also may have spilled over to competitors, a real world in which the menu of consortium activities changes over time with experimentation and learning.
|
From page 276... ...
, with c a decreasing function of its argument effective R&D with c' < 0, c" > 0. S is SEMATECH R&D, assumed a perfect substitute for internal company R&D, and ~ is the contribution of a dollar of shared, privately funded SEMATECH R&D to member-company effective R&D.
|
From page 277... ...
Thus, as Irwin and Klenow have argued, with pure information sharing (which is what is going on in this model) and SEMATECH effort a perfect substitute for internal R&D, the impact of a SEMATECH effort of size S will be to cut company effort by pS.
|
From page 278... ...
NXi + S _ 1 PQ N£ which is a familiar result from the Dasgupta-Stiglitz model (with S = 0~. Overall industry R&D (including SEMATECH projects)
|
From page 279... ...
SEMA TECH REVISITED APPENDIX B ORIGINAL IRWIN AND KLENOW SAMPLE MEMBERSHIP 279 COMPANY Broad Narrow Fabless Discrete Materials Other & Equipment Electronics Others SEMATECH DENSE-PAC MICROSYSTEMS INC LINEAR TECHNOLOGY CORP CYPRESS SEMICONDUCTOR CORP CHIPS&TECHNOLOGIES INC IMP INC OPTEK TECHNOLOGY INC DALLAS SEMICONDUCTOR CORP MAXIM INTEGRATED PRODUCTS LOGIC DEVICES INC AMERICAN MICROSYSTEMS APPLIED SOLAR ENERGY CORP TEXAS INSTRUMENTS INC AVANTEK INC BURR-BROWN CORP CODI CORP INTL RECTIFIER CORP DIONICS INC ELECTRONIC ARRAYS INC JETRONIC INDUSTRIES INC GENERAL SEMICONDUCTOR INDS SOLITRON DEVICES INC HYTEK MICROSYSTEMS INC BOURNS INC INTECH INC INTEGRATED DEVICE TECH INC INTERSIL INC INTERSIL INC NEW ALPHA INDUSTRIES INC UNITRODE CORP LSI LOGIC CORP MSI ELECTRONICS INC SEMTECH CORP NATIONAL SEMICONDUCTOR CORP MICRON TECHNOLOGY INC MICROSEMI CORP MICROWAVE SEMICONDUCTOR CORP MONOLITHIC MEMORIES INC MOSTEK CORP INTEL CORP ANALOG DEVICES ADVANCED MICRO DEVICES KYOCERA CORP -ADR SOLID STATE SCIENTIFIC RIPLEY CO INC SEEQ TE SEMICON INC SILICON SYSTEMS INC SILICONIX INC SILTEC CORP SUPERTEX INC TECCOR ELECTRONICS ALTERA CORP LATTICE SEMICONDUCTOR CORP XILINX INC ZILOG INC ATMEL CORP BROOKTREE CORP PMC-SIERRA INC INTEGRATED CIRCUIT SYSTEMS ADVANCED PHOTONIX INC -CLA TRANSTECTOR SYSTEMS INC VITESSE SEMICONDUCTOR CORP BKC SEMICONDUCTORS INC MICROCHIP TECHNOLOGY INC VLSI TECHNOLOGY INC XICOR INC APPIAN TECHNOLOGY INC ZITEL CORP EXAR CORP ELECTRONIC DEVICES INC PRECISION MONOLITHICS INC
|
From page 280... ...
280 SECURING THE FUTURE APPENDIX C ADDITIONAL FIRMS ADDED TO SAMPLE IN THIS STUDY COMPANY Broad Narrow Materials Other Fabless Discrete & Equipment Electronics Others SEMATECH ACTEL CORP AEROFLEX INC ALLIANCE SEMICONDUCTOR CORP ANADIGICS INC AUREAL INC BENCHMARQ MICROELECTRONICS CATALYST SEMICONDUCTOR INC CREE RESEARCH INC CYRIX CORP ELANTEC SEMICONDUCTOR INC ELECTRONIC DESIGNS INC ESS TECHNOLOGY INC GALILEO TECHNOLOGY LTD GATEFIELD CORP HEI INC IBIS TECHNOLOGY INC INFORMATION STORAGE DEVICES INTEGRATED SILICON SOLUTION LEVEL ONE COMMUNICATIONS INC MEMC ELECTRONIC MATRIALS INC MICREL INC MICRO LINEAR CORP MITSUBISHI ELEC CORP -ADR MRV COMMUNICATIONS INC OAK TECHNOLOGY INC OPTI INC QUALITY SEMICONDUCTOR INC RAMTRON INTERNATIONAL CORP REMEC INC SDL INC SIGMA DESIGNS INC SILICON STORAGE TECHNOLOGY SIMTEK CORP SIPEX CORP SMART MODULAR TECHNOLGS INC SPECTRUM SIGNAL PROCESSING STANDARD MICROSYSTEMS CORP STMICROELECTRONICS N V TELCOM SEMICONDUCTOR INC TOWER SEMICONDUCTOR LTD TRANSWITCH CORP TRIDENT MICROSYSTEMS INC UNIVERSAL DISPLAY CORP WHITE ELECTRIC DESIGNS CORP ZING TECHNOLOGIES INC ZORAN CORP
|
From page 281... ...
SEMA TECH REVISITED 281 APPENDIX D EQUIPMENT COMPANIES Company Name cusip SEMI/SEMATECH AG ASSOCIATES INC AMISTAR CORP AMTECH SYSTEMS INC APPLIED MATERIALS INC APPLIED SCI & TECH ASM INTERNATIONAL N V ASM LITHOGRAPHY HOLDING NV AS Y ST TECHNOLOGIES INC BE SEMICONDUCTOR INDUSTRIES BROOKS AUTOMATION INC BTU INTERNATIONAL INC CFM TECHNOLOGIES INC CHEMINEER INC CRYO-CELL INTERNATIONAL INC CVD EQUIPMENT CORP CYMER INC DT INDUSTRIES INC DTM CORP ELECTROGLAS INC EMCORE CORP ENGINEERED SYS & DEV CORP ETEC SYSTEMS INC FSI INTL INC GASONICS INTERNATIONAL CORP GCA CORP GENERAL SCANNING INC GENUS INC GERBER SCIENTIFIC INC HELISYS INC HI-RISE RECYCLING SYS INC IBIS TECHNOLOGY INC ICOS VISION SYSTEMS CORP NV INNOTECH INC INTEGRATED PROCESS EQ INTEVAC INC IONICS INC KULICKE & SOFFA INDUSTRIES LAM RESEARCH CORP MATERIALS RESEARCH MATTSON TECHNOLOGY INC MICRION CORP MRS TECHNOLOGY INC NOVELLUS SYSTEMS INC ONTRAK SYSTEMS INC OPAL INC PHOTRONICS INC PLASMA-THERM INC PRI AUTOMATION INC QC OPTICS INC QUAD SYSTEMS CORP RIMAGE CORP SEMITOOL INC SILICON VALLEY GROUP INC SONO-TEK CORP SPEEDFAM-IPEC INC SUBMICRON SYSTEMS CORP TEGAL CORP TOOLEX INTERNATIONAL NV TRIKON TECHNOLOGIES INC TURN TECHNOLOGY INC VARIAN SEMICONDUCTOR EQUIPMT VEECO INSTRUMENTS INC VITRONICS CORP WAVEMAT INC YIELDUP INTL CORP 107310 3153510 3233250 3822210 3823610 N0704510 N070591 1 04648X10 7332010 1 1434A10 5603210 12525K10 16381810 22889510 12660110 23257210 23333J10 23333L10 28532410 29084610 29286810 26922C10 30263310 36727810 36155620 37073710 37246110 37373010 42328210 42839610 45090910 B4923310 45766M10 45812K10 46114810 46221810 50124210 51280710 57668010 57722310 59479P10 55347610 67000810 68337410 68347410 71940510 72790010 69357H10 74693410 74730Q10 76672110 81690910 82706610 83548310 84770510 86431310 87900810 N8715N10 89618710 90021310 92220710 92241710 92850310 94356110 98583710
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.